VESSEL DIAMETER CHANGES DURING THE CARDIAC CYCLE

HEAN C. CHEN, VINOD PATEL, JUTTA WIEK, SALWAN M. RASSAM and EVA M. KOHNER London

SUMMARY likely to occur as a consequence of the pressure changes

Retinal vessel diameter, which is an important parameter during each cardiac cycle (pulsatility), autoregulation and in blood flow measurement, is affected by pulsation also as a result of vasomotion. 5 Spontaneous venous pulsa­ during the cardiac cycle and by vasomotion. This project tion of the central retinal vein is visible on direct ophthal­ studied these changes by analysing three monochromatic moscopy in a large proportion of the population; however, fundus photographs taken in eight arbitrary parts of the it is unknown whether smaller branches of the central ret­ cardiac cycle of 10 healthy subjects . It was found that the inal vein exhibit similar pulsations which would not be venous diameter decreased in early systole, increasing visible to the naked eye. Arterial blood flow being pulsa­ thereafter to a maximum level in early diastole and then tile in nature is likely to induce similar alterations in these declined towards end diastole. The maximum change of vessels. Autoregulation is the ability of a blood vessel to 4.82% (between early systole and early diastole) alter its diameter either to maintain constant blood flow in (p = 0.03) represents a 9.83% change in volumetric blood the face of changing perfusion pressure or to alter flow to flow. The arterial diameter peaked in mid-late systole, , satisfy local metabolic requirements.6.7 Vasomotion is the increasing by 3.46% (p = 0.01); this represents a blood spontaneous cyclical contraction and relaxation exhibited flow increase of 7.04%. Vasomotion led to changes of by some blood vessels; its presence has been demonstrated 3.71% and 2.61% in arteries and veins respectively. It is in the vasculature of the retina8 as well as of many other concluded that for accurate measurement of retinal blood tissues, although the degree to which vasomotion occurs flow, fundus photographs should be taken synchronised in the retina over a given period of time is unknown.9-13 with the electrocardiogram. Measurements of retinal blood flow that do not take these changes into account could be misleading. At rest, The retina is uniquely situated in allowing non-invasive pulsatility and vasomotion are likely to account for most direct visualisation of its circulatory system. This property of the vessel diameter changes, with autoregulation play­ means that the study of the retinal vasculature is of import­ ing an insignificant role. The aim of this study was to ance in furthering our understanding not only of retinal measure the variations in retinal vessel diameter in the vessel pathology but also of normal vessel physiology. healthy eye in the steady state. These vessel diameter The study of retinal blood flow has been important in changes may be altered in certain disease processes; the increasing our knowledge about the pathogenesis of vari­ pattern of alteration could help further elucidate the patho­ ous diseases, for example diabetic retinopathy, retinal genesis or aid in diagnosis. occlusive diseases and glaucoma.1-3 Volumetric retinal blood flowis dependent on both the velocity of blood flow SUBJECTS AND METHODOLOGY and the vessel diameter. With the advent of laser Doppler Ten healthy subjects were studied. The mean age was 34.6 velocimetry there now exists an accurate non-invasive years (range 29-57 years); there were 4 men and 6 women. method for directly measuring retinal erythrocyte velocity All had normal blood pressure and intraocular pressure in large retinal vessels.4 with the exception of one subject who had an intraocular Volumetric flowis calculated as the product of the mean pressure of 24 mmHg. Informed consent was obtained velocity of flow and the cross-sectional area of the vessel. from all subjects in accordance with the Helsinki Declara­ As the latter is dependent upon the square function of the tion. The camera used was a standard Zeiss 30° fundus diameter, even small changes will lead to significant camera (Zeiss, Oberkochen, Germany) using high-con­ changes in the flow rate. Vessel diameter fluctuations are trast Technical Pan film (Kodak, Rochester, NY, USA). The camera was fittedwith a 570 nm filterto obtain mono­ Correspondence to: H. C. Chen, Diabetic Retinopathy Unit, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, chromatic, red-free photographs which increases the res­ London Wl2 ONN, UK. olution of the vasculature.14

Eye (1994) 8, 97-103 © 1994 Royal College of Ophthalmologists 98 H. C. CHEN ET AL.

this delay period is a fraction of the duration of one cardiac cycle (R-R interval) which is dependent upon the subject's heart rate. The cardiac cycle was arbitrarily divided into eight parts and three photographs were taken in each of the parts. After the initial focusing requirements no further adjustments to the settings of the fundus camera were

p T made and the subjects were required to remain still R through the entire period. I 0/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 The first set of photographs was taken at 1I8th of the cardiac cycle from the R wave, to coincide with early sys­ I SYSTOLE DIASTOLE tole, through diastole, to 8/8th which coincides with the peak of the R wave. One cycle therefore extends between two R waves (O/8th to 8/8th) (Fig. 1). The photographs

Fig. 1. Division of the events of the cardiac cycle into eighths. were analysed using the Context Vision Digital Image Analysis system (Link6ping, Sweden).ls A high-resol­ The subjects were connected to an electrocardiogram ution TV camera scans the negatives and relays the image (ECG) monitor (American Optical, Bedford, Mass., USA) to the image analysis computer which digitises and dis­ with which the camera was synchronised through a time­ plays it on a 512 x 512 pixel array monitor. The measure­ delay switch. The time-delay switch was set to release the ment site on the vessel image is indicated by the placement camera shutter at any chosen period afterthe R wave of the of a cursor which cuts across the vessel perpendicularly. ECG to coincide with a particular part of the cardiac cycle; The computer then generates a transmittance profile of this site using the mean grey scale intensities from five loci distributed over 10 pixels on either side of the cursor. The background and peak intensity values of this profile are marked and from this the computer is able to determine its half-height; the width across the profileat its half-height is taken as the vessel diameter (Fig. 2).16 Three measure­ ments were made of each photograph. As three photo­ graphs were taken of each part of the cardiac cycle, there were nine readings in total. The average of these nine read­ ings was taken as the diameter for that part of the cardiac cycle. Care was taken to ensure that the measurements were made across the same point each time. Measure­ ments were made of a major vein and artery within 1 disc diameter of the optic disc. Comparisons were made of the vessel diameter at each part of the cardiac cycle with its value at the other seven parts. The overall variation through the cycle is expressed as the coefficientof variation of the cardiac cycle, which is the standard deviation expressed as a percentage of the mean diameter; this provides a measure of the degree of spread in the diameter value through the cycle. In order to evaluate changes in the diameter resulting from vasomotion, the degree of variation between the measurements of the three photographs in each part of the cardiac cycle was examined. Similarly the spread of values between the three photographs was expressed as the coefficient of variation resulting from vasomotion.

Table I. Venous diameter

Venous diameters (/--lm) SEM

118 143.04 7.98 Fig. 2. The negative of a fundus photograph after digitisation 2/8 145.20 8.17 by the Context Vision Image Analysis computer. It shows the 3/8 147.30 7.90 computer-generated transmittance profile of the retinal vessel 4/8 149.37 8.07 where it is marked by three perpendicular lines. The peak of this 5/8 149.93 8.84 6/8 148.83 7.97 profile and the background values are indicated by the place­ 7/8 147.12 7.54 ment of cursors; with these levels marked the computer is then 8/8 145.37 7.60 able to measure the width across the half-height of the profile. which is taken as the diameter. Mean values are shown for each eighth of the cardiac cycle (n = 10). VESSEL DIAMETER AND THE CARDIAC CYCLE 99

150

148

- e 146 -:::L ... Q) .- Q) i 144 is

142 ;r= 0 2 345 6 7 8 Eighths of the cardiac cycle

Fig. 3. Venous diameter means for each eighth of the cardiac cycle.

Statistical Methods The mean values for each part of the cardiac cycle are pre­ Analysis of variance was conducted on the data to ascer­ sented in Table I. The analysis of variance showed that tain whether significant intergroup differences existed. If there were differences between the different parts of the this was found to be significant at p<0.05 further analysis cardiac cycle (F = 4.98, p = 0.0002, d.f. = 7). There was was taken using a paired Student's t-test. Data are pre­ an initial reduction in the diameter at 1I8th (early systole) when compared with 0/8th (end diastole). This was fol­ sented as the mean ± standard error (unless stated other­ wise) and a probability of 0.05 was considered to be lowed by a gradual increase reaching a maximum level at 5/8th (early diastole) after which it started decreasing statistically significant. (Figs. 3, 4). The diameter measurement at l/8th was RESULTS smaller compared with the values at all other parts of the cardiac cycle, reaching statistical significance in the Venous Diameter differences with 3/8th (p = 0.03), 4/8th (p = 0.02), 5/8th The venous diameter ranged from lO5.9 /-lm to 212.8 /-lm. (p = 0.03) and 6/8th (p = 0.02). The same was true for the

5 *

4

3

2

o

-1

-2

-3 +-----.---�----�----T_--�----�----�--� o 2 3 4 5 6 7 8

* SEM Eighths of the cardiac cycle

Fig. 4. Percentage difference of each eighth of the cardiac cycle from 0/8th (venous). 100 H. C. CHEN ET AL.

Table IL Differences in venous diameter (%)

Jl8 2/8 3/8 4/8 5/8 6/8 7/8 8/8

1/8 2/8 +1.5% 3/8 +3.0%* +1.4%* 4/8 +4.4%* +2.9%* +1.4%* 5/8 +4.8%* +3.3%* +1.8% +0.4% 6/8 +4.0%* +2.5%* +1.0% -0.4% -0.7%* 7/8 +2.9% +1.3% -0.1% -1.5% -1.9% -1.1% 8/8 +1.6% +0.1% -1.3% -2.7%* -3.0% -2.3%* -1.2%

The percentage differences between parts of the cardiac cycle as shown. (The cardiac cycle is divided into eight parts and the '+' and '-' signs indicate the change relative to the preceding part of the cycle.) *p<0.05. measurement at 2/Sth, which also reached statistical sig­ cycle ranged from 1.24% to 4.64% with a mean of nificance with 3/Sth (p = 0.04), 4/Sth (p = 0.01), 5/Sth 2.32% ± 0.38%. (p = 0.02) and 6/Sth (p = 0.02). The measurements for 4/Sth and 6/Sth were significantly larger compared with Vasomotion O/Sth (p = 0.04 in both cases) (Table II). There is therefore Fluctuations in retinal vessel diameter may be due to pul­ an initial reduction of the vessel diameter in early systole satility (secondary to changes during the cardiac cycle), (represented by lISth and 2/Sth) relative to end diastole, vasomotion or autoregulation. By examining changes followed by a gradual increase reaching a maximum level between the three photographs taken in one part of the in early diastole, after which the diameter gradually cardiac cycle, the element of pulsatility is eliminated; decreases towards end diastole. The largest difference was similarly autoregulatory changes are unlikely to contrib­ between the values at lISth and 5/Sth, which represented a ute significantly to any changes in the period of time over change of 4.S2%; this corresponds to a 9.83% change in which the photographs were taken, given that the patients volumetric blood flow. The coefficients of variation for were in a relatively steady state. The difference between each individual subject through the cardiac cycle ranged the three photographs obtained from each part of the card­ from 0.91% to 4.50% with a mean of 2.1S% ± 0.51%. iac cycle (inter-photographic variation) was therefore taken as a measure of vasomotion. Although inter-photo­ Arterial Diameter graphic variation may result from methodological dis­ The arterial diameter ranged between 82.9!lm and crepancies, this possible source of variation was kept to a 150.4!lm (Table III). The analysis of variance showed minimum by maintaining a constant camera-to-eye dis­ that there were differences between the different parts of tance with no further adjustments to the camera settings the cardiac cycle (F = 2.36, p = 0.03, dJ. = 7). The dia­ once photography had commenced; it was also stressed to meter increased from 1I8th (early systole), reaching a peak the subjects the importance of steady fixation on a fixed at 3/8th (mid-late systole), and then gradually declined target with their contralateral eye and the photographs towards S/8th (end diastole) (Figs. 5, 6). The diameters at obtained were generally of a high standard. 2/8th and 3/Sth were both significantly greater than the The mean inter-photographic varlatlOns were value at lISth (p = 0.04, p = 0.01 respectively); the 3/8th 3.71 % ± 0.30% and 2.61% ± 0.24% for arteries and value was also significantly greater than that at 8/8th veins respectively; the corresponding coefficientsof vari­ (p = 0.04) (Table IV). There is therefore an initial increase ation were 1.90% ± 0.15% for arteries and in the arterial diameter reaching a peak in mid to late sys­ 1.36% ± 0.12% for veins. Since the coefficient of varia­ tole and decreasing thereafter towards diastole. The tion for the measurement technique (intra-photographic largest difference was the increase from lISth to 3/Sth, variation, which is the variation between the three which represented a change of 3.46%; this corresponds to measurements on any individual photograph) was only a change in volumetric blood flow of 7.04%. The coeffi­ 0.16% ± 0.02%, the inter-photographic values are cients of variation for each subject through the cardiac significant.

Table III. Arterial diameter DISCUSSION

Arterial diameter (11m) SEM This study has shown that both retinal arterial and venous diameters undergo change concomitant with the events of 118 115.61 6.36 2/8 118.85 6.58 the cardiac cycle. 3/8 119.34 6.40 The arterial change is perhaps to be expected as a con­ 4/8 118.95 6.61 sequence of the pulse wave entering the eye. The arterial 5/8 117.91 6.78 6/8 118.91 6.43 diameter reaches a maximum in mid systole, which is just 7/8 117.43 6.48 after the period of peak aortic pressure; subsequently, as 8/8 116.11 6.41 arterialpressure decays from its peak, the vessel diameter

Mean values are shown for each eighth of the cardiac cycle (n = 10). diminishes towards end diastole. 17 VESSEL DIAMETER AND THE CARDIAC CYCLE 101

Table IV. Differences in arterial diameter

1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8

118 2/8 +2.6%* 3/8 +3.2%* +0.4% 4/8 +2.9% +0. 1 o/c -0.3% 5/8 +2.0% -0.8% -1.2% -0.9clr 6/8 +2.1 % -0.6% -1.00/( -0.70/c +0.20/( 7/8 + 1.5% -1.2% -1.6% -1.30/( -0.4% -0.60/( 8/8 +0.40/( -2.3%* -2.7% -2.4% -1.50/0 -1.7% -1.1%

The percentage differences between parts of the cardiac cycle are shown. (The cardiac cycle is divided into eight parts and the '+' and '-' signs indicate the change relative to the preceding part of the cycle.) *p<0.05.

In early systole. as a result of a greater volume of blood venous and intraocular pressures. will not be subject to entering its vascular tree. the volume of the globe small alterations in the intraocular pressure; the artery is increases by up to 4%. which in turn increases the intra­ also a vessel with a thicker wall. ocular pressure by 1-2 mmHg, 18 The diameter of a retinal The results of this study are largely in agreement with vein. which is a thin-walled structure. is maintained by a the work by Delori et al.19 as to the times at which vessel balance of two main forces: the outwardly directed intra­ diameter is greatest and smallest. although they found the vascular pressure and the opposing intraocular pressure. change through the cycle to be smaller. They expressed These two pressures are normally very similar and small their results in terms of vessel diameter pulsatility (VDP) fluctuations in either may lead to large changes in the which was defined as the difference between the maximal transmural or distending pressure (intravascular pressure and minimal vessel diameter expressed as a percentage of minus the intraocular pressure); it is this distending prcs­ the pulse-averaged diameter; absolute values were not sure which is responsible for keeping the vessel opcn. The presented. They detected mean VDP values of increase in intraocular pressure during systole will lead to 2.3 ± 0.7% (SD) and 2.9 ± l.9% (SD) for arteries and a reduction in the distending pressure. This therefore veins respectively. These values are smaller than those of results in a passive reduction of the venous diameter and is the present study. which are 7.22 ± 3.9% (SD) and in accordance with the law of Laplace which states that the 6.45 ± 4.67% (SD) for arteries and veins respectively. distending pressure in a hollow object is proportional to its They did. however. employ a different mode of measure­ wall tension and radius. ment using scanning fundus reflectometry. The subsequent increase in venous diameter is due to a The magnitude of vasomotion in human retinal vessels combination of the systolic increase in blood flow reach­ has. to our knowledge. not previously been documented. ing the venous part of the vascular tree and the intraocular Riva et al.20 have previously shown a periodic variation in pressure returning to its diastolic level. The intra-arterial the blood flow of the optic nerve head. retina and choroid pressure. which is considerably higher than both the intra- at frequencies in the range of 2-4 per minute in miniature

120

- E -::L .. � - � i Q

o 2 3 4 5 6 7 8

Eighths of the cardiac cycle

Fig. 5. Arterial diameter means for each eighth of the cardiac CYcle. 102 H. C. CHEN ET AL.

5

* 4

- 3 �- t.lI 2 � C ..t.lI � - Q 0

-1

-2 0 2 3 4 5 6 7 8 * SEM Eighths of the cardiac cycle

Fig. 6. Percentage difference of each eighth of the cardiac cycle from 0/8th (arterial).

pigs which they attributed to vasomotion. This phenom­ with a diameter that is static. I I The vascular endothelium enon of spontaneous cyclical relaxation and contraction is plays an important role in the modulation of vasomotion well recognised in many tissues and may be altered by and is affected by conditions such as diabetes and hyper­ humoral, metabolic, physical and nervous stimuli." As ret­ tension; vasomotion is also thought to play a significant inal vessels do not possess autonomic innervation, the role in the pathogenesis of conditions such as unstable local factors are therefore likely to play an important role angina and variant angina.2I.22 in modulating the frequency and size of vasomotion; these Fig. 7 summarises the variations in diameter accounted same factors are also likely to be important in retinal for by the cardiac cycle, vasomotion and errors arising vessel autoregulation, which as earlier stated is a separate from the measurement technique. This emphasises the phenomenon. The degree of vasomotion may therefore be importance of coupling retinal photography with the EeG compromised in conditions where the local environment for accurate comparisons of retinal vessel diameters. We is altered. This alteration in vasomotive ability may be of have shown that an error of up to 7.04% and 9.83% varia­ fundamental importance in the underlying pathogenic tion in volumetric arterial and venous blood flowmeasure­ mechanisms - for example a vessel the diameter of which ments respectively may occur by not taking this into oscillates possesses a notably lower resistance than one account. It is therefore important, especially when serial

3

i-- c c i 2 • Arterial - cardiac cycle '� E3 Arterial - vasomotion '0 0 Venous - cardiac cycle - - C EI Venous vasomotion e � m Technique "; c U

0

Fig. 7. Diameter variation for arter.v and vein resulting from cardiac cycle changes, vasomotion and measurement technique. VESSEL DIAMETER AND THE CARDIAC CYCLE 103 photograpHs are taken, to study changes over time, that tuations in oxygen tension in the cat retina. Microvasc Res 1992;44:73-84. these photographs are taken synchronised with the ECG. 9. Nicholl PA, Webb RL. Vascular pattern and active vas­ Apart fromthe above, retinal vessel diameter changes in omotion as determiners of flow through minute vessels. the cardiac cycle may be altered in disease states and the Angiology 1955;6:291-310. pattern of variation may suggest possible pathogenic 10. D'Agrosa LS. Patterns of venous vasomotion in the bat mechanisms. In conditions in which the intravenous pres­ wing. Am J PhysioI1970;218:530-5. II. Colantuoni A, Bertuglia S, Intaglietta M. Quantitation of sure is raised, for example impending central retinal vein rhythmic diameter changes in arterial microcirculation. Am occlusion and raised intracranial pressure, there will be a 1 Physiol 1984;246: H508-17. diminution of the venous pulsatility. whereas a reduction 12. McAlpin RN. Contribution of dynamic vascular wall thick­ in the perfusion pressure as would occur, for example. ening to luminal narrowing during coronary arterial con­ striction. Circulation 1980;61:296-30 I. with a deficient carotid supply would lead to reduced 13. Intaglietta M. Endrich BA. Experimental and quantitative arterial pulsatility. Examining retinal vessel diameter analysis of microcirculatory water exchange. Acta Physiol changes in the cardiac cycle may therefore provide Scand Suppl 1979;463:59-66. another avenue for studying diseases which potentially 14. Delori FC, Gragoudas ES, Francisco R, Pruett RC. Mono­ compromise the circulation. chromatic ophthalmoscopy and fundus photography. Arch Ophthalmol 1977:95:861-8. Key words: Blood flow. Cardiac cycle. Fundus photographs. Pulsation. 15. Newsom RBS, Paul MS, Rassam 5MB. Jagoe R. Kohner Retinal vessel diameter. Vasomotion. EM. Retinal vessel measurement: comparison between observer and computer driven methods. Graefes Arch Clin REFERENCES Exp Ophthalmol 1992;230:221-5. 16. Brinchmann HO. Engvold O. Microphotometry of the blood 1. Patel V, Rassam S, Newsom R, Wiek J. Kohner E. Retinal column and the light streak on retinal vessels in fundus blood flow in diabetic retinopathy. BMJ 1992;305:678-83. photographs. Acta Ophthalmol (Cophenh) 1986;95:861-8. 2. Grunwald JE, Riva CEo Stone RA, Keates EU. Petrig BL. 17. Ganong WF. The heart as a pump. In: Review of medical Retinal autoregulation in open-angle glaucoma. Ophthal­ physiology. Los Altos: Lange Medical Publications. 1985: mology 1984;91:1690-4. 459-69. 3. Fujino T, Curtin VT, Norton EW D. Experimental central ret­ 18. AIm A. Ocular circulation. In: Hart WM Jr, editor. Adler's inal vein occlusion: a comparison of intraocular and extra­ physiology of the eye. St Louis: Mosby Year Book. 1992: ocular occlusion. Arch Ophthalmol 1969;81 :395-406. 198-227. 4. Riva CE, Grunwald JE, Sinclair SH. O'Keefe K. fundus 19. Delori FC, Fitch KA. Variation of retinal vessel diameter camera based retinal LDY. Appl Optics 1981;20: 117-20. with cardiac cycle [abstract]. Invest Ophthalmol Vis Sci 5. Wise G, Dollery C, Henkind P. Physiologic principles. In: 1988;29: 178. The retinal circulation. New York: Harper & Row, 1971: 20. Riva CE, Pournaras CJ. Poitry-Yamate CL, Petrig BL. 83-117. Rhythmic changes in velocity, volume and flow of blood in 6. Johnson Pc. Review of previous studies and current theories the optic nerve head tissue. Microvasc Res 1990;40:36-45. of autoregulation. Circulation Res 1964; 14/15(Suppl.l): 21. Harrison DG. Kurz MA, Quillen lE, Sellke FW, Mugge A. 2-9. Normal and pathophysiologic considerations of endothelial 7. Guyton AC. Local control of blood flow by the tissues. and regulation of vascular tone and their relevance to nitrate nervous and humoral regulation. In: Textbook of medical therapy. Am J Cardiol 1992;70:B 11-7. physiology. Philadelphia: Saunders, 1981;232-45. 22. Pepine CJ. Daily life ischemia and nitrate therapy. Am J 8. Braun RD, Linsenmeier RA, Yancey CM. Spontaneous fluc- Cardiol I 992;70:B54-63.